WashPo Opioid Dataset – large data viz with R

At this meetup we have an interesting presentation and training from Winston Saunders on data that is too large to fit into memory! We look forward to having you!

Abstract:
In 2019 the Washington Post successfully pursued a FOIA request for records of opiod shipments from the DEA’s Automated Reporting System (ARCOS) spanning the height of the prescription opiod epidemic from 2006 to 2012. These data were subsequently published for 179 Million drug transactions ranging from mg to kilograms in size. While R’s powerful analytical and visualization capabilities are, in principle, great for understanding the structure of the distribution network that enabled the epidemic, given the 78GB size of the dataset, efficient analysis required developing techniques, including data parsing and preparation of “summary” data sets, to dig deeply and thoroughly into the data. I’ll share some of the techniques used to wrangle the raw data into more digestible summaries as well as the tools developed and the summaries themselves. We’ll also explore how these summary data sets, coupled with mapping (ggmap), network analysis (network, igraph), and Shiny, can be used to reveal the magnitude the opiod epidemic and the network architecture, from manufacturer, to distributor, to consumer, that enabled it. To wrap up, we’ll access one of the summary datasets and walk through a quick network visualization or mapping hands-on exercise. [To do the ggmaps mapping exercise you’ll need your own API key from google]

Biography:
Winston is a data hackster with a day job. Though he first learned R in the Johns Hopkins Coursera pre-tidyverse he now enthusiastically embraces %>% and, nine times in ten, chooses slice(1:10) over [1:10, ]. He’s explored ML, NLP, AI, image processing, web & twitter bots, kaggle competitions, and lots of other applications in R. He selected his current project out of a balanced sense of responsibility and mischief.

If you are wanting to have your own meetup to discuss the cool ways you are using R, reach out to Ellis Hughes via Meetup.com!